The fibroblast growth factor (FGF) signaling pathway has been implicated in a variety of developmental processes of vertebrates. It has been shown that FGF has potent mesoderm-inducing properties in Xenopus animal caps (Kimelman and Kirschner, 1987; Slack et al., 1987) and is required for posterior mesoderm formation and convergent extension movements (Amaya et al., 1991, 1993). FGF is also known as one of the posteriorizing signals (Cox and Hemmati-Brivanlou, 1995; Kengaku and Okamoto, 1995; Lamb and Harland, 1995; Taira et al., 1997; Gamse and Sive, 2000), which include retinoic acid and members of Wnt family (Durston et al., 1989; McGrew et al., 1997; Gamse and Sive, 2000; Yamaguchi, 2001). In addition, FGF is required for the formation of the neuroectoderm, including the prospective brain region (Launay et al., 1996; Hongo et al., 1999). The mechanisms of these distinct roles of FGF in mesoderm formation, neuralization, and posteriorization need to be clarified. One possible explanation may be functional differences between various members of the FGF family (Hardcastle et al., 2000; Yang et al., 2002), as well as FGF receptors and coreceptors (Szebenyi and Fallon, 1999; Simons and Horowitz, 2001).
Recently, nou-darake (ndk), a gene encoding a transmembrane protein related to the FGF receptor (FGFR) but lacking a cytoplasmic kinase domain, has been isolated in the planarian Dugesia japonica (Cebria et al., 2002). Loss of function of ndk by RNA interference (RNAi) results in the induction of ectopic brain tissues throughout the body. Furthermore, the ectopic brain formation is suppressed by the inhibition of planarian FGFR homologs (FGFR1 and FGFR2), and NDK inhibits FGF signaling in Xenopus embryos, suggesting that NDK may modulate FGF signaling in stem cells to restrict brain tissues to the head region of planarians. In vertebrates, FGFR-like 1 (FGFRL1) is known as a transmembrane protein related to FGFR in the ectodomain but lacks a cytoplasmic kinase domain as well as NDK (Wiedemann and Trueb, 2000, 2001; Kim et al., 2001; Sleeman et al., 2001). In mice and humans, FGFRL1 is expressed in various tissues, including cartilage and the pancreas, but their expression and roles in early development have not been investigated. To explore the role of FGFRL1/ndk in Xenopus embryos, we isolated the Xenopus ortholog of FGFRL1/ndk (XFGFRL1/Xndk) and analyzed the spatiotemporal expression patterns of XFGFRL1/Xndk in Xenopus development.
Cloning and Structure of XFGFRL1 cDNA
To isolate the cDNA fragment of XenopusFGFRL1/ndk, a Xenopus EST clone that showed high similarity to the human FGFRL1 cDNA sequence with an E-value of 1.1E-21 was selected using the TIGER Xenopus laevis Gene Index (XGI). A 241-bp fragment was amplified from a cDNA library of Xenopus stage 17/18 whole-embryo using primers constructed from the EST sequence. The deduced amino acid sequence encoded in this fragment shows 74% identity to that of human FGFRL1 in the second Ig-like domain. cDNA library screening was performed using the 241-bp fragment as a probe. An isolated cDNA clone (1,874 bp long) contained 310 bp of 5′ untranslated region (UTR), 1,452 bp of open reading frame (ORF) encoding 484 amino acid residues and 112 bp of 3′ UTR. This predicted protein is highly related to human FGFRL1 as well as planarian NDK. Unlike these proteins, which contain three FGFR-related Ig-like domains, planarian NDK does not have conserved cysteine residues in the third Ig-like domain or a typical transmembrane domain. Their cytoplasmic regions lack a kinase domain characteristic of FGF receptors and do not have significant homology to any other known sequences. We noticed that the reported C-terminal sequence of mouse FGFRL1 (Sleeman et al., 2001; Wiedemann and Trueb, 2001) does not match those of the human and Xenopus sequences and that an additional reading frame of the mouse sequence does match to the human and Xenopus sequences, suggesting that the reported mouse sequences need to be reexamined. Based on the similarity to the overall structure of human FGFRL1, the gene we isolated is likely to be the Xenopus ortholog of FGFRL1, referred to as XFGFRL1 or Xndk. Xenopus FGFRL1 shows 66% or 17% identity to that of human or planarian, respectively, over the entire region (Fig. 1).
Expression Patterns of XFGFRL1 in Early Development
The temporal expression pattern of XFGFRL1 in early Xenopus development was analyzed by reverse transcription-polymerase chain reaction (RT-PCR; Fig. 2). XFGFRL1 expression was first detectable at the early neurula stage (stage 13/14) and maintained at later stages. The spatial and temporal expression pattern was analyzed by whole-mount in situ hybridization. At the late gastrula stage (stage 12), XFGFRL1 expression was widely detected in the anterior region when cleared with benzyl benzoate/benzyl alcohol (Fig. 3A). At the early neurula stage, expression of XFGFRL1 became prominent in the anterior head region with a clear posterior boundary (Fig. 3B). To examine the expression domain precisely, stained embryos were sectioned and stained with DAPI (4′,6-diamidine-2-phenylidole) to distinguish germ layers by the differences in nuclear density. This examination by sectioning indicated that XFGFRL1 expression in the anterior region is localized to the prechordal plate, anterior endoderm, and archenteron roof (Fig. 3C) and also to anterior part of the paraxial and lateral mesoderm (Fig. 3D).
As XFGFRL1 is structurally related to FGFR and may modulate FGF signaling, as shown in the functional analysis of NDK in Xenopus embryos (Cebria et al., 2002), we observed that expression domains of XFGFRL1 at later stages resemble those of Xenopus, chick, and mouse FGF8 (Heikinheimo et al., 1994; Ohuchi et al., 1994; Crossley and Martin, 1995; Mahmood et al., 1995; Christen and Slack, 1997; Crossley et al., 2001) in the anterior neural ridge (ANR), midbrain–hindbrain boundary (MHB) region, forebrain, otic vesicles, somites, and visceral arches as described below. We therefore carried out whole-mount in situ hybridization for both XFGFRL1 and XFGF8 at stage 17/18 and later. As neurulation proceeded, the mesodermal expression of XFGFRL1 gene expanded to the posterior and ventral regions (Fig. 3E, left and middle panels), and additional expression was detected at the ANR (Fig. 3E, left panel), while expression of XFGF8 was detected at the ANR and also in the ventrolateral region in a crescent shape (Fig. 3E, right panel; indicated by an arrowhead). Section examination showed that the expression of both XFGFRL1 (Fig. 3F, upper panels) and XFGF8 (Fig. 3F,H, lower panels) at the ANR was detected primarily in its sensorial layer. While the ventrolateral crescent-shaped expression of XFGF8 is mainly in the anterior sensorial layer of the ectoderm (indicated by arrowheads in Fig. 3F,G, lower panels), XFGFRL1 is more broadly expressed in the mesoderm adjacent to the crescent-shaped XFGF8 expression domain (Fig. 3E, left; 3F,G, upper panels). The expression region of XFGFRL1 in the overall anterior head region is shown in a cleared embryo (Fig. 3E, middle panel) and includes the prechordal plate, anterior endoderm, and archenteron roof (Fig. 3G,H, upper panels), similar to the early neurula (Fig. 3B–D). At the neural tube stage (stage 22), XFGFRL1 expression was detected in the eye vesicles and the forebrain (Fig. 3I, upper left and right panels). In addition, the expression in the MHB region was manifested when stained embryos were cleared (Fig. 3I, upper middle panel). At this stage, XFGF8 expression was detected at the rostral midline of the forebrain, MHB, and otic placodes (Fig. 3I, lower panels), and the crescent-shaped expression was separated into two stripes (indicated by arrowheads in Fig. 3I, lower panels).
During the tail bud stages (stages 26 to 33/34), additional expression of XFGFRL1 was detected in the visceral arches and the otic vesicles (Fig. 4A,B, left panels). XFGFRL1 was coexpressed with XFGF8 in the MHB, forebrain, visceral arches, and somites at stage 26 and later, although the expression level of XFGF8 in the somites decreases at later stages (Fig. 4A–D,G,H). In the otic vesicles, XFGFRL1 and XFGF8 appeared to be expressed differently in the ventral and anterior region, respectively, of the vesicles (Fig. 4B; see below). XFGFRL1 expression was also detected in the pineal gland (Fig. 4B, left panel; 4C, left and middle panels; 4E, upper left panel), while XFGF8 expression was detected in the pronephros and pronephric duct at stage 26 (Fig. 4A, right panel), and probably at the boundary between prosomeres 2 and 3 at stages 26 to 33/34 (indicated by arrowheads in Fig. 4A,B, right panel) according to the observation in the chick embryo (Crossley et al., 2001).
Horizontal sections showed that XFGFRL1 expression was detected in the overall forebrain, while XFGF8 expression was localized in the rostral midline of the forebrain and in anterior head mesenchyme (Fig. 4D) and that XFGFRL1 and XFGF8 were coexpressed in the visceral pouches of pharyngeal endoderm (Fig. 4G). Figure 4E shows that XFGFRL1 expression in the eye is localized in the lens epithelial cells, transitional zone, and ciliary margin (Fig. 4E, upper panels), while XFGF8 expression in the eye is localized in the anterior ciliary margin and commissural plate (Fig. 4E, lower panels) as has been reported in chick embryos (Crossley et al., 2001). The expression of XFGFRL1 and XFGF8 in ventral and anterior region, respectively, of the otic vesicles was confirmed by examination of a cross-section for XFGFRL1 (Fig. 4F, upper panel) and a horizontal section for XFGF8 (Fig. 4F, lower panel). Expression of both XFGFRL1 and XFGF8 in the somites at stage 26 was localized around the nucleus (Fig. 4H, indicated by arrows).
Conserved Expression Patterns of FGFRL1/ndk in Xenopus and Planarian
We have identified the Xenopus ortholog of FGFRL1/ndk, analyzed the expression patterns in Xenopus development, and found that XFGFRL1 expression is detected in an overall anterior mesendoderm at neurula stages (Fig. 3). This is the first report to show that the vertebrate FGFRL1 gene is expressed in the head region at the early embryonic stage, and, remarkably, this anterior expression pattern is very much akin to that of ndk (Cebria et al., 2002), suggesting that FGFRL1/ndk expression in the head region is conserved in evolution between planarians and Xenopus. Furthermore, ndk expression in the head region of planarians is detected in both brain and nonbrain tissues, similar to the neuroectodermal and mesendodermal expression of XFGFRL1 in Xenopus. Evolutionarily conserved expression patterns in the head region between planarians and vertebrates have been shown in Otx genes (Umesono et al., 1999). The planarian orthologs of Otx, DjotxA, and DjotxB are expressed in the head region, including brain tissue and mesenchymal cells, while the Otx2 gene in vertebrates is expressed in the anterior neuroectoderm corresponding to fore- and midbrain as well as the head organizer region corresponding to the prechordal plate (Blitz and Cho, 1995; Simeone and Acampora, 2001). However, differing from XFGFRL1, Otx2 is not expressed in the anterior ventrolateral region of mesoderm and endoderm.
The structural features of FGFRL1 indicate that this protein may interact with FGF signaling. We have shown that NDK inhibits Xbra expression induced by FGF2 in animal caps, and NDK inhibits Xbra expression in whole embryos (Cebria et al., 2002). Consistent with this, FGF2 has been shown to bind to mouse FGFRL1 (also referred to as FGFR5; Sleeman et al., 2001). FGFs are thought to be involved in neuralization of the ectoderm and posteriorization of neural tissue, based on the observations that overexpression of dominant negative XFGFR1 (XFD) or XFGFR4 (FGFR4Δ) inhibits the expression of posterior neural marker genes (Kroll and Amaya, 1996; Bang et al., 1997; Taira et al., 1997) as well as neuralization (Launay et al., 1996; Hongo et al., 1999). Thus, anterior expression of XFGFRL1 at neural stages suggests a possibility that XFGFRL1 inhibits posteriorizing effects of FGFs and may maintain anterior head region of embryos.
As has been shown previously, loss of function of ndk by RNAi leads to induction of ectopic brain tissues in the trunk region (Cebria et al., 2002). Based on this non–cell autonomous role of ndk in planarian brain formation, XFGFRL1 may also have an indirect role in vertebrate brain development.
Synexpression Group of FGFRL1 and FGF8
We found that, at tail bud stages, XFGFRL1 is coexpressed with XFGF8 in many regions, including the MHB, midline region of forebrain, visceral arches, and somites (Fig. 4). Members of the Sprouty family (Minowada et al., 1999; Furthauer et al., 2001), sef (for “similar expression of FGF genes;” Furthauer et al., 2002; Tsang et al., 2002), Isthmin (Pera et al., 2002), and FGF8 are postulated as the “synexpression group” that share a similar, spatiotemporal expression pattern and may be involved in the same biological process. Indeed, it has been shown that the intracellular Sprouty protein and the putative transmembrane protein Sef act as negative-feedback regulators of FGF signaling (Minowada et al., 1999; Furthauer et al., 2001, 2002; Tsang et al., 2002). This synexpression group also includes other members of FGF family, FGF17 and FGF18, in mouse and zebrafish embryos (Maruoka et al., 1998; Reifers et al., 2000). Thus, our data indicate that FGFRL1 belongs to the synexpression group of FGF8, and can function as a negative regulator for FGF signaling, similar to Sprouty and Sef.
Studies with zebrafish, chick, and mouse embryos have shown that FGF8 is an essential signaling factor that mediates the organizing activity of the MHB (Crossley et al., 1996; Liu and Joyner, 2001; Rhinn and Brand, 2001) and forebrain development (Meyers et al., 1998; Shanmugalingam et al., 2000; Fukuchi-Shimogori and Grove, 2001). Because XFGFRL1 and XFGF8 are coexpressed in these two regions, XFGFRL1 may be involved in the formation of MHB and forebrain by regulating FGF8 signaling. In the regions where XFGF8 is not expressed, XFGFRL1 may be coexpressed with other FGFs. For example, FGF1 and FGF2 are expressed in both epithelium and transitional zone of the lens (McAvoy et al., 1999) where XFGFRL1 is expressed (Fig. 4E) and have been shown to stimulate proliferation of lens epithelial cells and differentiation of lens fiber cells (McAvoy and Chamberlain, 1989; Schulz et al., 1993). Recently, it has been shown that the lens-specific Maf gene is expressed in the lens epithelial cells in Xenopus embryos and regulated by FGF for lens fiber differentiation (Ishibashi and Yasuda, 2001). These overlapping expression domains of XFGFRL1 and other FGFs suggest that XFGFRL1 interacts with FGF signaling in these regions as well.
In summary, identification of XFGFRL1/Xndk and precise analysis of the expression patterns in Xenopus development have suggested that anterior expression patterns of FGFRL1/ndk are conserved in evolution between planarians and frogs, and that XFGFRL1 belongs to the FGF8 synexpression group and is possibly involved in development of various regions, including the brain through the modulation of FGF signaling.
Cloning of XFGFRL1 cDNA
A 241-bp fragment was PCR-amplified from a λZAP II-based cDNA library of Xenopus neurula (stage 17/18; constructed by J. Shinga and M. Taira) by using the primers 5′-AGTCATTGCTAGGCCAT*TG-3′ and 5′-GACTTCTACTTTGTACGTCG-3′, which were designed from a Xenopus expressed sequence tag sequence (GenBank accession no. BJ088492) (T* was substituted to G in cDNA we isolated) that showed high similarity to the human FGFRL1 cDNA. The Xenopus neurula cDNA library was screened by plaque hybridization using a 32P-labeled probe produced from the 241-bp fragment. Positive phage clones were converted to plasmids by ZAPing out (Stratagene). Both strands were sequenced by using deleted constructs made by restriction digestion with the fluorescence-labeled T3 (5′-AATTAACCCTCACTAAAG-3′) and T7 (5′-TAATACGACTCACTATAGGG-3′) primers and a LiCor 4200 sequencer (Aloka) or using an internal primer, 5′-CACATTCTCACGTGGAGGCC-3′ and an ABI 310 sequencer (Perkin-Elmer).
Manipulations of Xenopus Embryos
Xenopus embryos were fertilized in vitro, dejellied, and reared in 0.1 × Steinberg's solution (Kay and Peng, 1991). Embryos were staged according to Nieuwkoop and Faber (1967).
RNA Extraction and RT-PCR
Total RNA was extracted from embryos as previously described (Osada et al., 2003). RT-PCR analysis was performed using the primers 5′-AGTCATTGCTAGGCCAGTG-3′ and 5′-GACTTCTACTTTGTACGTCG-3′, and cDNA was transcribed with SUPERSCRIPT II Reverse Transcriptase (GIBCO BRL).
Whole-Mount In Situ Hybridization
Antisense digoxigenin (DIG) -RNA probes were transcribed from SK(-)XFGFRL1 (linearized with EcoRI) or XFGF8 (linearized with HindIII) (Endo et al., 2000) by using T7 or T3 RNA polymerase, respectively, and DIG-RNA Labeling mix (Boehringer Mannheim). Whole-mount in situ hybridization of albino embryos was performed by using an automated in situ hybridization system (AIH-101, Aloka) according to the method described by Harland (1991). Stained embryos were paraffin-sectioned at 15 μm, and the nuclei were stained with DAPI as described (Shibata et al., 2001). To obtain strong signals of DAPI-stained nuclei on sections, washing out of the yellowish color after Bouin fixation of BM purple-stained embryos was done within 16 hr.
We thank H. Ide for XFGF8, S. Ohmori for preparation of RNAs from embryos at various stages, and T. Shingyoji for supporting S. Hayashi.